On Tue, May 25, 2010 at 3:48 PM, Roan Kattouw <roan.kattouw(a)gmail.com> wrote:
Also note that you wouldn't technically be
parsing, just preprocessing
on the home wiki, which is certain to be less expensive (how much less
I don't know)
This is a good point.
and that you'd be doing this on some wiki anyway,
only the overhead involved in HTTP, the API and initializing the
parser is relevant; the actual cost of the operation is not, because
you're doing it someplace either way (of course this only applies
intra-WMF, not to external clients).
External clients are what I'm worried about. It's a nonissue for
intra-Wikimedia use, but if external clients start using a lot of CPU
by using Wikimedia servers for parsing, I expect them to get shut
down, and no one wants that.
On Tue, May 25, 2010 at 4:09 PM, Marco Schuster
Why not mix it? Take other templates etc. from the
source wiki and set magic
stuff like time / contentlang to target wiki values.
That's what I suggested, basically. Use templates from the foreign
wiki, but do the actual parsing locally, so you get local values for
magic words and so on.